I’m excited to share my new Streamlit app that uses AI to separate vocals and instruments from music. With this app, you can listen to your favorite songs with just the vocals or instruments (bass, drums and others), or even remove them completely to create a karaoke-style experience!
You can check out the app deployed here.
The app uses a pretrained model called Hybrid Spectrogram and Waveform Source Separation from Facebook’s htdemucs library to do the separation.
I want also to mention also that I’m using this custom component for audio recorder.
I would love to hear your feedback on the app and any ideas for improvements. Feel free to leave a comment on this post or open an issue on the GitHub repository if you run into any issues.
Thanks for checking it out, and happy music separating!